Misleading Stats.

Yesterday, I ended up teaching a half day of school before taking students out for a 2.5 hour walk on the tundra.  I gave students a chance to catch up with their work, I looked for a unit on misleading statistics for students who were up to date.

While searching I found this wonderful document titled "How to lie, cheat, manipulate, and mislead using statistics and graphical displays" from UCSD.

This presentation is wonderful because it begins with definitions of statistics and such before moving on to explain different types of bias found in interpreting the data.  Each type of bias begins with a certain situation being given such as the recording temperatures by one buoy in the ocean around San Diego. 

The person takes time to explain the sample and the population parts of the situation.  It goes on to explain that if the data is applied to certain situations, it becomes a biased sampling and why it is called that.  From here, the presentation moves on to explain four types of biased sampling - area, self selection, leading question, and social desirability, each with the appropriate examples.

The next step is to look at the different types of data analysis used to manipulate information such as poor analysis, averages, and best of all, graphical displays which make manipulating data so much easier.  I love the way data is taken and using different ways of displaying graphically, people could come to the wrong conclusion.  The last couple of pages shares good graphical displays.

Combine this with a great article from statistics how to and you have a good introduction to the topic because it shows graphs which look wonderful but are totally misleading.  One example shows a newspaper comparing its circulation to another one.  At first glance, it appears the first one has double the circulation of the second but if you look at the actual scale, there is only a difference of about 40,000 readers or about 10%.

The misleading graphs are divided into missing the baseline, incomplete data, numbers not adding up correctly, two Y axis, and just reading it wrong, each with one or more examples.  It is great because the author of this included a written description of the problem and for one  included what the graph should look like.

Check both sites out and let me know what you think.  I had fun finding these and I plan to use them in class when we spend a couple weeks looking at statistics and probability.  I hope you enjoyed it as much as I.  Have a good day.